Method and system for rapidly and dynamically measuring permeability coefficient of unsaturated soil

Abstract

The invention discloses a method and a system for rapidly and dynamically measuring the permeability coefficient of unsaturated soil. The method comprises the following steps: acquiring resistivity data, dielectric constant data and temperature data of soil in a target area; collecting the data according to a unified time resolution to obtain observation data; constructing a soil heterogeneous unit feature vector based on the collected data; dividing the feature vectors of the heterogeneous units of the soil to generate an initial permeability coefficient field; constructing a rock physical constitutive model; fusing observation data and a hydrodynamic model, combining with a rock physical constitutive model, dynamically updating a permeability coefficient field, and performing optimization by applying physical constraints; and finally outputting the updated permeability coefficient spatial-temporal distribution and uncertainty quantification result. According to the invention, the moisture change in the unsaturated soil can be monitored in real time, the moisture distribution condition of the soil can be accurately evaluated, and geological disaster risks such as leakage and landslide of the soil can be found in advance.

Description

Technical Field

[0001] The present invention relates to the technical fields of geotechnical engineering and hydrogeology, and specifically to a method and system for rapidly and dynamically measuring the permeability coefficient of unsaturated soil. Background Art

[0002] The permeability coefficient of unsaturated soil is a core parameter describing the soil water migration ability. Traditional measurement methods such as the instantaneous profile method and the centrifugation method have problems such as high cost, low efficiency, destructiveness, and insufficient resolution. Existing physical technologies such as ERT and GPR can indirectly reflect soil characteristics, but the inversion accuracy of a single data source is limited, and there is a lack of dynamic update ability, making it impossible to achieve real-time measurement with high spatio-temporal resolution. In addition, the seepage characteristics of heterogeneous soils are affected by the coupling of multiple factors such as porosity, temperature, and volumetric water content. Existing static models do not consider the dynamic coupling effect of multiple factors and also lack a quantitative evaluation of the reliability of inversion results. Therefore, it is necessary to integrate multi-source geophysical data and hydrogeological measurement data to solve the problem of spatio-temporal dynamic high-precision measurement of the permeability coefficient of unsaturated soil. Summary of the Invention

[0003] Aiming at the deficiencies of the prior art, the present invention provides a method and system for rapidly and dynamically measuring the permeability coefficient of unsaturated soil, aiming to achieve rapid and high-precision dynamic measurement of the permeability coefficient of unsaturated soil through multi-source physical data fusion, fuzzy C-means clustering, and ensemble Kalman filter dynamic assimilation technology, thereby solving the limitations of traditional measurement methods, improving the measurement speed and accuracy of the permeability coefficient, and realizing dynamic monitoring of soil water characteristics.

[0004] To achieve the above object, the present invention provides the following technical solution: A method for rapidly and dynamically measuring the permeability coefficient of unsaturated soil, comprising the following steps: Step S1: Obtain resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; collect the resistivity data, dielectric constant data, and temperature data of the soil in the target area at a unified time resolution to obtain observation data; Step S2: Construct a characteristic vector of soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data; Step S3: Based on the fuzzy C-means clustering algorithm, divide the characteristic vector of soil heterogeneous units to identify and classify different soil characteristics, and at the same time generate an initial permeability coefficient field of the soil in the target area; Step S4: Construct a rock physics constitutive model and establish a non-linear relationship between the permeability coefficient in the resistivity data, dielectric constant data, and the initial permeability coefficient field of the soil in the target area and soil physical parameters; Step S5: Based on the permeability coefficient, soil temperature, and soil physical parameters, generate an initial ensemble through the adaptive ensemble Kalman filtering algorithm. Input the initial ensemble into the hydrodynamic model to generate a prediction ensemble. Combine with the rock physics constitutive model, map the prediction ensemble to the observation space, and obtain the predicted values of soil resistivity data, dielectric constant data, and permeability coefficient at different times for the soil in the target area. Dynamically update the prediction ensemble based on the predicted values and the observed data; Step S6: Based on the updated prediction ensemble, output the spatio-temporal distribution of the permeability coefficient in the prediction ensemble and the uncertainty quantification results.

[0005] Further, the specific process of Step S3 is as follows: Establish the objective function of the fuzzy C-means clustering algorithm , expressed as: (1); In the formula, represents the total number of characteristic vectors of soil heterogeneous units; represents the number of clustering categories; represents the membership degree of the th characteristic vector of the soil heterogeneous unit belonging to the th category; represents the fuzzy index; represents the th characteristic vector of the soil heterogeneous unit; represents the clustering center of the th category; represents the spatial regularization weight; represents the neighborhood of the th characteristic vector of the soil heterogeneous unit; represents the th characteristic vector of the soil heterogeneous unit and the th membership degree between the clustering centers of the th characteristic vector of the soil heterogeneous unit and the th membership degree between the clustering centers of the th category; represents the th characteristic vector of the soil heterogeneous unit and the index of other characteristic vectors of the soil heterogeneous unit in the neighborhood; By alternately optimizing the membership degree and the clustering center until convergence, map the clustering result to the initial permeability coefficient field of the soil in the target area , expressed as: (2); In the formula, represents the initial permeability coefficient field at the th position of the soil in the target area; Indicates the typical value of the permeability coefficient of the class of clusters.

[0006] Furthermore, the specific process of step S4: respectively establish the resistivity data of the soil in the target area , dielectric constant data and the initial permeability coefficient field of the permeability coefficient in the soil physical parameters including porosity , saturation and volumetric water content , expressed as: (3); (4); (5); In the formula, represents the cementation factor; , respectively represent the porosity index and the saturation index; represents the volumetric water content; represents a constant related to the particle shape; represents the saturation index; represents the density of water; Construct a multi-parameter joint inversion equation: Combine equations (3), (4) and (5) to form a closed equation system, and input the resistivity data of the soil in the target area , dielectric constant data , and solve the porosity and saturation by Newton-Raphson method iteration, and then substitute the solved and into equation (5) to determine the calculation model of the permeability coefficient .

[0007] Furthermore, the specific process of step S5 is: According to the soil porosity , volumetric water content , permeability coefficient and soil temperature of the soil in the target area, construct an initial state vector , , containing soil dynamic state variables ( ) and static parameters ( ): (6); In the formula, Indicates transpose; Based on the initial state vector Generate an initial set, including set members, and the set members are obtained by adding random perturbations to the initial state vector ; Run the hydrodynamic model to predict the states of all set members. Specifically: Use the hydrological model operator to predict the state of each set member at the next time step, forming a prediction set, expressed as: (7); In the formula, represents the th set member at time; represents the th set member at time; is the hydrological model operator; is the time step; Combine the rock physics constitutive model to map the prediction set to the observation space, and obtain the soil resistivity data , dielectric constant data , permeability coefficient predicted values of , expressed as: (8); In the formula, is the rock physics model; represents the set prediction generated by the hydrological model operator; Compare the predicted value with the observed data and calculate the Kalman gain matrix , and update the prediction set through the Kalman gain matrix, expressed as: (9); (10); In the formula, is the prediction error covariance matrix; is the observation error; , are the prediction sets before and after update, respectively; represents transpose of; Feed back the updated prediction set to the hydrological model operator for the simulation of the next time step; Meanwhile, covariance inflation and localization are introduced during the inversion process of the adaptive ensemble Kalman filtering algorithm, and physical constraints are applied; the process of the adaptive ensemble Kalman filtering algorithm is repeated until the distributions of the predicted values and the observed values are consistent; the convergence criterion is set as the predicted value and the observed data with a Nash-Sutcliffe efficiency coefficient NSE > 0.8 between their distributions, and the change in three consecutive iterations < 1%.

[0008] Furthermore, the specific process of step S6 is as follows: Using Kriging interpolation, the discrete permeability coefficients in the updated prediction ensemble are mapped to a continuous spatio-temporal grid; and through statistical indicators and probability distributions, the uncertainty of the permeability coefficient is quantified, and the probability that the permeability coefficient exceeds the threshold is calculated to indicate potential seepage risks.

[0009] Furthermore, the uncertainty entropy of the permeability coefficient is calculated by equation (11), and then the probability that the permeability coefficient exceeds the threshold is calculated by equation (12), expressed as: (11); (12); In the formula, represents the uncertainty entropy of the permeability coefficient at time; represents the probability that the permeability coefficient belongs to the category at time; represents the probability that is greater than the permeability coefficient threshold, i.e., the preset critical value ; represents the permeability coefficient of the th sample; represents the total number of categories of the permeability coefficient; the sample is a member of the updated prediction ensemble.

[0010] Furthermore, before constructing the characteristic vector of the soil heterogeneous unit based on the collected resistivity data, dielectric constant data, and temperature data, the resistivity data, dielectric constant data, and temperature data are preprocessed, and the preprocessing includes denoising and normalization.

[0011] A rapid dynamic measurement system for the permeability coefficient of unsaturated soil includes: A data acquisition module for obtaining the resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; the resistivity data, dielectric constant data, and temperature data of the soil in the target area are collected at a unified time resolution to obtain observed data; The feature vector construction module is used to construct the feature vectors of soil heterogeneous units based on the collected resistivity data, dielectric constant data, and temperature data; The partitioning module is used to partition the feature vectors of soil heterogeneous units based on the fuzzy C-means clustering algorithm to identify and classify different soil properties, and simultaneously generate the initial permeability coefficient field of the soil in the target area; The model construction module is used to construct a petrophysical constitutive model and establish the non-linear relationship between the permeability coefficient and soil physical parameters in the resistivity data, dielectric constant data, and initial permeability coefficient field of the soil in the target area; The update and optimization module is used to generate an initial ensemble based on the permeability coefficient, soil temperature, and soil physical parameters, input the initial ensemble into the hydrodynamic model to generate a prediction ensemble, jointly with the petrophysical constitutive model, map the prediction ensemble to the observation space to obtain the predicted values of the soil resistivity data, dielectric constant data, and permeability coefficient at different times of the soil in the target area, and dynamically update the prediction ensemble based on the predicted values and the observed data; The output module is used to output the spatio-temporal distribution of the permeability coefficient and the uncertainty quantification results in the updated prediction ensemble.

[0012] An electronic device includes a processor, a memory, and a bus. The processor and the memory are connected through the bus. Among them, the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a method for rapidly and dynamically measuring the permeability coefficient of unsaturated soil.

[0013] A non-volatile computer storage medium stores computer-executable instructions, and the computer-executable instructions execute a method for rapidly and dynamically measuring the permeability coefficient of unsaturated soil.

[0014] Compared with the existing technologies, the present invention has the following beneficial effects:

[0015] (1) Through the collaborative innovation of multi-source data fusion, FCM clustering, dynamic assimilation algorithm, and ensemble statistics, the present invention realizes the high-precision, high-efficiency, and high-resolution measurement of the permeability coefficient of unsaturated soil, and simultaneously provides the functions of uncertainty quantification and risk warning, significantly improving the dynamic characterization ability of the seepage process of complex heterogeneous soil, and providing scientific theoretical support for geological disaster prevention and agricultural water conservation.

[0016] (2) Supported by the dynamic assimilation algorithm, the present invention can update the unsaturated permeability coefficient at the hourly level, quickly capture the dynamic changes of transient hydrological events, overcome the deficiency that the traditional method can only estimate the permeability coefficient statically and isolatedly, and significantly improve the timeliness of monitoring and prediction.

[0017] (3) Through the multi-source data fusion algorithm, the spatial resolution of the present invention reaches the centimeter level, which can finely depict soil heterogeneity and solve the problem of local detail loss caused by traditional homogeneous assumptions and isolated inversion.

[0018] (4) The present invention quantifies uncertainty through set statistics, provides a confidence interval for risk decision-making, and makes up for the limitation of the lack of credibility evaluation in a single solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a flowchart of the method of the present invention.

[0020] Figure 2 It is a schematic diagram of data collection of the present invention.

[0021] Figure 3 It is a schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0022] As Figure 1 shown, the present invention provides a technical solution: a method for rapidly and dynamically measuring the permeability coefficient of unsaturated soil, including the following steps:

[0023] Step S1: Obtain the resistivity data , dielectric constant data and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; the resistivity data, dielectric constant data and temperature data of the soil in the target area are collected at a unified time resolution to obtain observation data .

[0024] Among them, the resistivity data , dielectric constant data and temperature data of the soil in the target area are obtained by using the high-density resistivity method (ERT), the ground penetrating radar method (GPR) and soil temperature sensors respectively; as Figure 2 shown, Figure 2 in, 1 is the soil in the target area (unsaturated soil), 2 is the high-density electrical survey line, 3 is the ground penetrating radar antenna, 4 is the soil temperature sensor, 5 is the main unit of the high-density electrical method instrument, 6 is the receiver of the soil temperature sensor, 7 is the main unit of the ground penetrating radar, 8 is the data processing PC, and the resistivity data , dielectric constant data and temperature data of the soil in the target area (unsaturated soil) are collected through the high-density electrical survey line 2, the ground penetrating radar antenna 3 and the soil temperature sensor 4 respectively, received through the main unit 5 of the high-density electrical method instrument, the receiver 6 of the soil temperature sensor and the main unit 7 of the ground penetrating radar respectively, and finally transmitted to the data processing PC to complete the collection.

[0025] Step S2: Based on the collected resistivity data , dielectric constant data and temperature data construct the characteristic vector of soil heterogeneous units , where \(T\) represents transpose.

[0026] Among them, before constructing the characteristic vector of soil heterogeneous units based on the collected resistivity data , dielectric constant data and temperature data , the resistivity data , dielectric constant data and temperature data can be pre - processed (to eliminate the dimensional differences between them). The pre - processing includes denoising and normalization: , , ; \(\widetilde{\rho}_i\) represents the pre - processed resistivity data ; \(i\) represents the \(i\) - th resistivity data; \(\overline{\rho}\) represents the mean value of the resistivity data; \(\sigma_{\rho}\) represents the standard deviation of the resistivity data; \(\widetilde{\varepsilon}_j\) represents the pre - processed dielectric constant data ; \(j\) represents the \(j\) - th dielectric constant data ; \(\overline{\varepsilon}\) represents the mean value of the dielectric constant data; \(\sigma_{\varepsilon}\) represents the standard deviation of the dielectric constant data; \(\widetilde{T}_k\) represents the pre - processed temperature data ; \(k\) represents the \(k\) - th temperature data ; \(\overline{T}\) represents the mean value of the temperature data; \(\sigma_T\) represents the standard deviation of the temperature data.

[0027] Step S3: Based on the fuzzy C - means (FCM) clustering algorithm, partition the characteristic vector of soil heterogeneous units to identify and classify different soil properties, and at the same time generate the initial permeability coefficient field of the soil in the target area .

[0028] The specific process of Step S3 is: Establish the objective function of the fuzzy C - means (FCM) clustering algorithm , expressed as: (1); In the formula, represents the total number of characteristic vectors of soil heterogeneous units; represents the number of clustering categories; represents the th characteristic vector of soil heterogeneous unit belonging to the th class membership degree; represents the fuzzy index, usually taken as 2; represents the th characteristic vector of soil heterogeneous unit; represents the th class clustering center; represents the spatial regularization weight; represents the th neighborhood of the characteristic vector of soil heterogeneous unit; represents the th characteristic vector of soil heterogeneous unit and the th class clustering center between the membership degree; represents the th characteristic vector of soil heterogeneous unit and the th class clustering center between the membership degree, represents the th characteristic vector of soil heterogeneous unit in the neighborhood of the index of other characteristic vectors of soil heterogeneous unit.

[0029] The objective function of the fuzzy C - means clustering algorithm designed by the present invention innovatively introduces a spatial weight matrix, that is, the second term in formula (1), to force the membership degrees of adjacent units to be similar and suppress the "checkerboard" artifacts.

[0030] By alternately optimizing the membership degree and the clustering center until convergence, the clustering result is mapped to the initial permeability coefficient field of the target area soil , expressed as: (2); In the formula, represents the initial permeability coefficient field of the target area soil at the th position; represents the typical permeability coefficient values of the th class clustering, and these values can be determined based on prior knowledge, experimental data or experience.

[0031] Step S4: Construct a petrophysical constitutive model to establish the resistivity data of the target area soil , dielectric constant data and the initial permeability coefficient field in the permeability coefficient and soil physical parameters (porosity , saturation and volumetric water content ) of the non - linear relationship

[0032] The specific process of step S4: First, according to Archie's law (Equation (3)), Topp's formula (Equation (4)) and Kozeny - Carman equation (Equation (5)), establish the resistivity data of the soil in the target area , dielectric constant data and the initial permeability coefficient field in the permeability coefficient and soil physical parameters (porosity , saturation and volumetric water content ) of the quantitative relationship, expressed as: (3); (4); (5); In the formula, represents the cementation factor (sand , clay ); , respectively represent the porosity index and saturation index (usually , ); represents the volumetric water content; represents a constant related to particle shape; represents the saturation index; represents the density of water

[0033] Construct a multi - parameter joint inversion equation: Combine Equation (3), Equation (4) and Equation (5) to form a closed system of equations, and input the resistivity data of the soil in the target area , dielectric constant data , and solve for the porosity and saturation by the Newton - Raphson method iteration. Then substitute the solved and into Equation (5) to determine the calculation model of the permeability coefficient . The calculation model of the permeability coefficient can quickly and conveniently calculate the soil permeability coefficient

[0034] Step S5: Based on the permeability coefficient , soil temperature and soil physical parameters (porosity , saturation and volumetric water content ), an initial ensemble is generated through the Adaptive Ensemble Kalman Filtering (AEnKF) algorithm. The initial ensemble is input into the hydrodynamic model to generate a prediction ensemble. Combining with the petrophysical constitutive model (Equations (3) to (5)), the prediction ensemble is mapped to the observation space to obtain the soil resistivity data at different times in the target area , dielectric constant data , permeability coefficient prediction values , based on the prediction values and the observed data to dynamically update the prediction ensemble.

[0035] The specific process of step S5 is as follows: According to the soil porosity , volumetric water content , permeability coefficient and soil temperature obtained from the above steps, an initial state vector containing soil dynamic state variables ( , , ) and static parameters ( ) is constructed: : (6); In the formula, represents the transpose.

[0036] Based on the initial state vector to generate an initial ensemble to characterize the uncertainty, including ensemble members (usually = 50 ∼ 200), and the ensemble member is obtained by adding a random perturbation to the initial state vector .

[0037] Run the hydrodynamic model to perform state predictions for all ensemble members. Specifically, use the hydrodynamic model operator (such as Hydrus-3D software) to predict the state of each ensemble member at the next time step, forming a prediction ensemble, expressed as: (7); In the formula, represents the state of the th ensemble member at time; represents the state of the th ensemble member at time; is the hydrodynamic model operator (such as Hydrus-3D software); is the time step, generally taking the time resolution of the observed data.

[0038] Combining the joint rock physics constitutive model (Equations (3) to (5)), the prediction ensemble is mapped to the observation space to obtain the soil resistivity data at different times in the target area , dielectric constant data , permeability coefficient prediction values , expressed as: (8); where is the rock physics constitutive model, referring to Archie's law (Equation (3)), Topp's formula (Equation (4)), and Kozeny-Carman equation (Equation (5)); represents the ensemble prediction generated by the hydrological model operator.

[0039] Compare the predicted values with the observed data and calculate the Kalman gain matrix , and update the prediction ensemble through the Kalman gain matrix, expressed as: (9); (10); where is the prediction error covariance matrix; is the observation error; , are the prediction ensembles before and after update, respectively; represents transpose.

[0040] Feed the updated prediction ensemble back to the hydrological model operator (such as Hydrus-3D software) for simulation of the next time step.

[0041] At the same time, covariance inflation (Inflation factor = 1.05~1.3) and localization (using the Gaspari-Cohn function) are introduced during the inversion process of the adaptive ensemble Kalman filter (AEnKF) algorithm to alleviate the ensemble sampling error. And physical constraints (upper and lower limits of the permeability coefficient, which can be set to (10 −10 ~10 -3 m / s)) are applied to avoid physically unreasonable values (such as negative permeability coefficients, etc.) and suppress artifacts. Repeat the above process until the predicted values and the observed data are consistently distributed, and the convergence criterion is set as a global index, that is, the predicted values and the observed data The Nash efficiency coefficient NSE between the distributions is > 0.8, and the change in three consecutive iterations is < 1%; the predicted value and the observed data The Nash efficiency coefficient NSE between the distributions can be expressed as: ; represents the th predicted value; represents the th observed data; represents the average value of the observed data; represents the number of observed data or predicted values.

[0042] Step S6: Based on the updated prediction set, output the spatio-temporal distribution of the permeability coefficient in the prediction set and the uncertainty quantification result.

[0043] The specific process of Step S6 is as follows: Using the Kriging interpolation method, map the discrete permeability coefficients in the updated prediction set to a continuous spatio-temporal grid; and quantify the uncertainty of the permeability coefficient through statistical indicators and probability distributions, and calculate the probability that the permeability coefficient exceeds the threshold to prompt potential seepage risks.

[0044] Among them, calculate the uncertainty entropy of the permeability coefficient through Equation (11), and then calculate the probability that the permeability coefficient exceeds the threshold through Equation (12), expressed as: (11); (12); In the formula, represents the uncertainty entropy of the permeability coefficient at the moment; represents the probability that the permeability coefficient at the moment belongs to the category ; represents the probability that it is greater than the permeability coefficient threshold (preset critical value) ; represents the permeability coefficient of the th sample (set member in the updated prediction set); represents the total number of all possible states or categories of the permeability coefficient.

[0045] As Figure 3 shown, a rapid dynamic measurement system for the permeability coefficient of unsaturated soil, characterized in that it includes:

[0046] A data acquisition module, which is used to obtain the resistivity data, dielectric constant data and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; the resistivity data, dielectric constant data and temperature data of the soil in the target area are collected at a unified time resolution to obtain observation data.

[0047] A feature vector construction module, which is used to construct the characteristic vector of soil heterogeneous units based on the collected resistivity data, dielectric constant data and temperature data.

[0048] A partitioning module, which is used to partition the characteristic vector of soil heterogeneous units based on the fuzzy C-means clustering algorithm to identify and classify different soil properties, and at the same time generate the initial permeability coefficient field of the soil in the target area.

[0049] A model construction module, which is used to construct a petrophysical constitutive model and establish the non-linear relationship between the permeability coefficient in the resistivity data, dielectric constant data and the initial permeability coefficient field of the soil in the target area and the soil physical parameters.

[0050] An update and optimization module, which is used to generate an initial ensemble based on the permeability coefficient, soil temperature and soil physical parameters, input the initial ensemble into the hydrodynamic model to generate a prediction ensemble, combine with the petrophysical constitutive model, map the prediction ensemble to the observation space to obtain the predicted values of the soil resistivity data, dielectric constant data and permeability coefficient at different times of the soil in the target area, and dynamically update the prediction ensemble based on the predicted values and the observation data.

[0051] An output module, which is used to output the spatio-temporal distribution of the permeability coefficient in the prediction ensemble and the uncertainty quantification result based on the updated prediction ensemble.

[0052] An electronic device, including a processor, a memory and a bus, the processor and the memory are connected through the bus, wherein the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a method for rapidly and dynamically determining the permeability coefficient of unsaturated soil.

[0053] A non-volatile computer storage medium stores computer-executable instructions, and the computer-executable instructions execute a method for rapidly and dynamically determining the permeability coefficient of unsaturated soil.

[0054] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A rapid dynamic measurement method for the permeability coefficient of unsaturated soil, characterized in that, It includes the following steps: Step S1: Obtain the resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; collect the resistivity data, dielectric constant data, and temperature data of the soil in the target area at a unified time resolution to obtain observation data; Step S2: Construct a soil heterogeneous unit feature vector based on the collected resistivity data, dielectric constant data, and temperature data; Step S3: Based on the fuzzy C-means clustering algorithm, divide the soil heterogeneous unit feature vector to identify and classify different soil properties, and at the same time generate the initial permeability coefficient field of the soil in the target area; Step S4: Construct a petrophysical constitutive model to establish the non-linear relationship between the resistivity data, dielectric constant data of the soil in the target area, and the permeability coefficient in the initial permeability coefficient field and soil physical parameters; Step S5: Based on the permeability coefficient, soil temperature, and soil physical parameters, generate an initial ensemble through the adaptive ensemble Kalman filtering algorithm, input the initial ensemble into the hydrodynamics model to generate a prediction ensemble, combine the petrophysical constitutive model, map the prediction ensemble to the observation space, obtain the predicted values of the soil resistivity data, dielectric constant data, and permeability coefficient at different times of the soil in the target area, and dynamically update the prediction ensemble based on the predicted values and observation data; Step S6: Based on the updated prediction ensemble, output the spatio-temporal distribution and uncertainty quantification results of the permeability coefficient in the prediction ensemble.

2. The rapid dynamic measurement method for the permeability coefficient of unsaturated soil according to claim 1, wherein: The specific process of step S3 is as follows: Establish the objective function of the fuzzy C-means clustering algorithm , which is expressed as: （1）； In the formula, represents the total number of characteristic vectors of soil heterogeneous units; represents the number of clustering categories; represents the th membership degree of the characteristic vector of the soil heterogeneous unit belonging to the th class; represents the fuzzy index; represents the th characteristic vector of the soil heterogeneous unit; represents the clustering center of the th class; represents the spatial regularization weight; represents the th neighborhood of the characteristic vector of the soil heterogeneous unit; represents the th membership degree between the characteristic vector of the soil heterogeneous unit and the clustering center of the th class; represents the th membership degree between the characteristic vector of the soil heterogeneous unit and the clustering center of the th class, represents the index of other characteristic vectors of soil heterogeneous units in the th neighborhood of the characteristic vector of the soil heterogeneous unit; By alternately optimizing the membership degree and the cluster center After convergence, map the clustering result to the initial permeability coefficient field of the soil in the target area , which is expressed as: （2）； In the formula, represents the initial permeability coefficient field at the th position of the soil in the target area; represents the typical value of the permeability coefficient of the th type of clustering.

3. A rapid dynamic measurement method for the permeability coefficient of unsaturated soil according to claim 2, characterized in that: Specific process of step S4: Establish resistivity data of soil in the target area , dielectric constant data and an initial permeability coefficient field in which the permeability coefficient has a quantitative relationship with soil physical parameters including porosity , saturation and volumetric water content , expressed as: （3）； （4）； （5）； wherein, represents the cementation factor; , respectively represent the porosity index and the saturation index; represents the volumetric water content; represents a constant related to the particle shape; represents the saturation index; represents the density of water; Construct a multi-parameter joint inversion equation: Combine Equation (3), Equation (4), and Equation (5) to form a closed equation system, and input the resistivity data of the soil in the target area , dielectric constant data , and iteratively solve for porosity and saturation by the Newton-Raphson method. Then substitute the solved and into Equation (5) to determine the calculation model of the permeability coefficient .

4. A method for rapidly and dynamically measuring the permeability coefficient of unsaturated soil according to claim 3, characterized in that: The specific process of Step S5 is as follows: Based on the soil porosity , volumetric water content , permeability coefficient and soil temperature of the soil in the target area, construct an initial state vector , , containing soil dynamic state variables ( ) and static parameters : （6）； In the formula, represents transpose; Based on the initial state vector Generate an initial set, including set members, and the set members are obtained by adding random perturbations to the initial state vector ; Run the hydrodynamics model to perform state prediction on all ensemble members. Specifically: use the hydro-model operator to predict the state of each ensemble member at the next time step to form a prediction ensemble, expressed as: （7）； In the formula, represents the state of the th set member at moment; represents the state of the th set member at moment; is a hydrological model operator; is the time step; The combined rock physics constitutive model maps the prediction set to the observation space to obtain the soil resistivity data of the target area at different times , dielectric constant data , permeability coefficient predicted values of , expressed as: （8）； In the formula, is the petrophysical model; represents the ensemble prediction generated by the hydrological model operator; Comparison predicted value with the observed data and calculate the Kalman gain matrix , update the prediction set through the Kalman gain matrix, expressed as: （9）； （10）； In the formula, is the prediction error covariance matrix; is the observation error; , are the prediction sets before and after update, respectively; denotes is the transpose of Feed the updated prediction ensemble back to the hydro-model operator for simulation at the next time step; Meanwhile, covariance inflation and localization are introduced during the inversion process of the adaptive ensemble Kalman filtering algorithm, and physical constraints are applied; the process of the adaptive ensemble Kalman filtering algorithm is repeated until the distributions of the predicted values and the observed values are consistent; the convergence criterion is set as the predicted value and the observed data with a Nash-Sutcliffe efficiency coefficient NSE > 0.8 between their distributions, and the change in three consecutive iterations < 1%.

5. A method for rapidly and dynamically measuring the permeability coefficient of unsaturated soil according to claim 4, characterized in that: The specific process of Step S6 is as follows: Adopt Kriging interpolation method to map the discrete permeability coefficient in the updated prediction ensemble to a continuous spatio-temporal grid; and quantify the uncertainty of the permeability coefficient through statistical indicators and probability distributions, and calculate the probability that the permeability coefficient exceeds the threshold to prompt potential seepage risks.

6. A rapid dynamic measurement method for the permeability coefficient of unsaturated soil according to claim 5, characterized in that: Calculate the uncertainty entropy of the permeability coefficient through Equation (11), and then calculate the probability that the permeability coefficient exceeds the threshold through Equation (12), expressed as: （11）； （12）； In the formula, represents the permeability coefficient at the moment of the uncertainty entropy; represents at the moment the permeability coefficient belongs to the category with the probability; represents the probability that it is greater than the permeability coefficient threshold, i.e., the preset critical value ; represents the permeability coefficient of the th sample; represents the total number of categories of the permeability coefficient; the sample is a set member in the updated prediction set.

7. A rapid dynamic measuring method for the permeability coefficient of unsaturated soil according to claim 6, characterized in that: Before constructing the soil heterogeneous unit feature vector based on the collected resistivity data, dielectric constant data, and temperature data, preprocess the resistivity data, dielectric constant data, and temperature data. The preprocessing includes denoising and normalization.

8. A rapid dynamic measurement system for the permeability coefficient of unsaturated soil, characterized in that, It includes: A data acquisition module for obtaining the resistivity data, dielectric constant data, and temperature data of the soil in the target area; the soil in the target area is unsaturated soil; Collect the resistivity data, dielectric constant data, and temperature data of the soil in the target area at a unified time resolution to obtain observation data; A feature vector construction module for constructing a soil heterogeneous unit feature vector based on the collected resistivity data, dielectric constant data, and temperature data; A partitioning module, which is used to partition the eigenvectors of soil heterogeneous units based on the fuzzy C-means clustering algorithm to identify and classify different soil properties, and simultaneously generate an initial permeability coefficient field of the soil in the target area; A model construction module, which is used to construct a rock physics constitutive model and establish a non-linear relationship between the resistivity data, dielectric constant data of the soil in the target area, and the permeability coefficient in the initial permeability coefficient field and soil physical parameters; An update and optimization module, which is used to generate an initial ensemble based on the permeability coefficient, soil temperature, and soil physical parameters, input the initial ensemble into a hydrodynamic model to generate a prediction ensemble, jointly with the rock physics constitutive model, map the prediction ensemble to the observation space to obtain predicted values of the soil resistivity data, dielectric constant data, and permeability coefficient of the soil in the target area at different times, and dynamically update the prediction ensemble based on the predicted values and the observed data; An output module, which is used to output the spatio-temporal distribution of the permeability coefficient and the uncertainty quantification result in the prediction ensemble based on the updated prediction ensemble.

9. An electronic device, characterized in that, It includes a processor, a memory, and a bus. The processor and the memory are connected through the bus. Among them, the memory is used to store a set of program codes, and the processor is used to call the program codes stored in the memory to execute a rapid dynamic measurement method for the permeability coefficient of unsaturated soil according to any one of claims 1-7.

10. A non-volatile computer storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions execute a rapid dynamic measurement method for the permeability coefficient of unsaturated soil according to any one of claims 1-7.

Images